Proteomic and genomic characterization of a yeast model for Ogden syndrome

Abstract Naa10 is an Nα‐terminal acetyltransferase that, in a complex with its auxiliary subunit Naa15, co‐translationally acetylates the α‐amino group of newly synthetized proteins as they emerge from the ribosome. Roughly 40–50% of the human proteome is acetylated by Naa10, rendering this an enzyme one of the most broad substrate ranges known. Recently, we reported an X‐linked disorder of infancy, Ogden syndrome, in two families harbouring a c.109 T > C (p.Ser37Pro) variant in NAA10. In the present study we performed in‐depth characterization of a yeast model of Ogden syndrome. Stress tests and proteomic analyses suggest that the S37P mutation disrupts Naa10 function and reduces cellular fitness during heat shock, possibly owing to dysregulation of chaperone expression and accumulation. Microarray and RNA‐seq revealed a pseudo‐diploid gene expression profile in ΔNaa10 cells, probably responsible for a mating defect. In conclusion, the data presented here further support the disruptive nature of the S37P/Ogden mutation and identify affected cellular processes potentially contributing to the severe phenotype seen in Ogden syndrome. Data are available via GEO under identifier GSE86482 or with ProteomeXchange under identifier PXD004923. © 2016 The Authors. Yeast published by John Wiley & Sons, Ltd.


Introduction
N α -terminal acetylation (NTA) is one of the most prevalent protein modifications known with 80-90% of the human proteome being acetylated at the N-terminus Lange et al., 2014). NTA is catalysed by N α -terminal acetyltransferases (NATs) that transfer the acetyl moiety from acetyl-CoA to the α-amino group of their protein substrates. Despite its discovery more than 50 years ago, the functional consequences of this modification are not fully understood. Early studies indicated that NTA can modulate protein-protein interaction (Deakin et al., 1980;Scott et al., 2011;Nazmi et al., 2012). This feature seems to be especially relevant for proteolytic processes where NTA increases the affinity of ubiquitin ligases towards their substrates. As such, NTA creates degradation signals (Ac/N-degrons) that can be recognized by the ubiquitin ligase Doa10 and Not4 that promote ubiquitination and proteasome-mediated degradation in conjunction with Ubc6 and Ubc7 (Hwang et al., 2010;Shemorry et al., 2013;Park et al., 2015). In other cases, NTA has been shown to increase protein stability by blocking the access for N-terminal ubiquitination, thereby protecting proteins from proteasomal degradation Kuo et al., 2004;Aksnes et al., 2015).
To date six NATs have been discovered in humans (NatA-NatF) with a distinct substrate specificity . The best studied NAT is NatA, acetylating around 40-50% of the human and yeast proteome Starheim et al., 2012). NatA is composed of Naa10 (Ard1p in yeast) and its auxiliary subunit Naa15 (Nat1p). Naa15 is believed to anchor the complex to the ribosome to enable co-translational NTA as soon as a newly synthetized peptide emerges from the exit tunnel (Gautschi et al., 2003;Raue et al., 2007;). Bound to Naa15, the catalytic subunit Naa10 acetylates the α-amino group of proteins starting with alanine, serine, glycine, threonine, valine and cysteine, after the initiator methionine has been cleaved by methionine-aminopeptidases Arnold et al., 1999;Polevoda et al., 1999). In the monomeric form, Naa10 displays a different substrate specificity, post-translationally acetylating proteins starting with aspartate or glutamate Foyn et al., 2013) Structural analyses showed that the reason for this shift in substrate specificities are conformational changes induced by the binding to Naa15 (Liszczak et al., 2013). Lysine acetyltransferase activity has also been attributed to Naa10; however, structural and biochemical data strongly suggests that Naa10 does not have the properties to facilitate direct ε-acetylation of lysine sidechains (Liszczak et al., 2013;Magin et al., 2016).
Recently, we have identified a c.109 T > C (p. Ser37Pro) variant in NAA10 in two families with a lethal X-linked disorder of infancy which we named Ogden syndrome in honour of where the first family lives (in Ogden, Utah). The disorder comprises a distinct combination of an aged appearance, craniofacial anomalies, hypotonia, global developmental delays, cryptorchidism and cardiac arrhythmias (Rope et al., 2011). Functional analysis of this mutation revealed a decreased enzymatic activity of the S37P mutant in vitro and in primary cells derived from the patients as well as impaired complex formation of Naa10 and Naa15 (Rope et al., 2011;Myklebust et al., 2015;. Furthermore, we identified affected proteins which were less NTA in patient cells, one of which is THOC7 (THO complex subunit 7 homologue). Notably, the reduced acetylation of THOC7 was associated with a decreased stability of the protein, possibly contributing to Ogden syndrome (Myklebust et al., 2015).
Since the first description of Ogden syndrome in two multiplex unrelated families, additional variants in Naa10 have been identified in other probands with non-syndromic intellectual disabilities accompanied by postnatal growth failure and skeletal anomalies (Rauch et al., 2012;Popp et al., 2014;Saunier et al., 2016), one family with two brothers with syndromic intellectual disability with long QT, a prologation of the depolarization and repolarization interval of the ventricles of the heart (Casey et al., 2015) and one multiplex family with Lenz microphthalmia syndrome, characterized by microphthalmia or anophthalmia, developmental delay, intellectual disability, skeletal abnormalities and malformations of teeth, fingers and toes (Esmailpour et al., 2014). The phenotypic differences between all cases are fairly distinct, and to date there has been no unifying explanation for this, beyond just genetic background differences.
Here, we expanded on a prior study  by using Saccharomyces cerevisiae to study the impact of Naa10 disruption in several different physiologic situations and by conducting genomic and proteomic assays with an emphasis on the S37P/Ogden mutation.

Yeast strains
Derivatives of parental stain W303-1A (leu2-3,112 trp1-1 can1-100 ura3-1 ade2-1 his3-11,15) were used to generate strains with variations of the endogenous yNAA10 and yNAA15 loci. To introduce the human S37P mutation in yeast (YG36), first the homologue position was identified as Serine 39 by sequence alignment (Fig. 1A). yNAA10 was amplified from W303-1A using the primers 5′-GTA GAA TTC GCC GCC ATG CCT ATT AAT ATT CGC AG and 5′-CAT GAA TTC CCT ACC GAA TTA GCA CTG CAG T and cloned into pBEVY-U (Addgene stock #51230). The yS39P mutation was introduced using the QuikChange Multi Site-Directed Mutagenesis Kit (Agilent) and the 5′-ATG TAT CAT ATT CTC CCG TGG CCG GAG GCT T primer. yNAA10- Figure 1. Characterization of strains used in this study. (A) Sequence alignment of human and Saccharomyces cerevisiae Naa10. Protein sequences were aligned using PROMALS3D (Pei et al., 2008) and the crystal structure of Naa10 from Saccharomyces pombe (PDB structure 4KVX) (Liszczak et al., 2013) was used for secondary structure prediction. The alignment matrix between human and yeast Naa10 reached a 49.5% identity score. The homologous position to hNaa10 S37P (yeast Serine 39) is marked with an arrow. (B) RT-PCR characterization of the human replacement strains using primers designed to amplify regions internal to the indicated transcripts. The products of these reactions are shown, in the presence and absence of reverse transcriptase. (C) Western blot analyses with antibodies for the human and yeast Naa10 and Naa15 along with yGAPDH as a loading control for the strains where the yNaa10 has been modified, (D) for the humanized strains, expressing the human proteins from the endogenous locus and (E) for the strains overexpressing the human proteins from plasmids 21 A yeast model for Ogden syndrome S39P was then PCR amplified using 5′-GGG AAA  CCT AAA TAC ATA CGA TCA AGC TCC AAA  ATA AAA CTT CGT CAA CCA TGC CTA TTA  ATA TTC GCA GAG CG and 5′-TGT GAA GAA  GCC TGG ATG AAA ATA TAC TAC GTT TAT  ATA GGT TGA TTT AAT TAT ACA ATG ATA  TCA TTT ACG CCT TGC primers and the resulting PCR product was transformed into the deletion stain YG10 (ΔyNAA10::URA3-15) using the lithium acetate technique (Amberg et al., 2005). Transformants were selected using 5-FOA and verified by sequencing of the yNAA10 locus followed by western blot analysis (Fig. 1B). The double deletion strain (YG67: ΔNaa10/ Δ Naa15) was generated by crossing YG3 and YG10 to form a diploid cell (grown on SC -Leu,Ura ). Once the cells diploidized, tetrad dissections were performed to isolate a haploid double knockout. Candidates were selected on SC -Leu,Ura plates and screened for mating type and auxotrophies by replica plating.
To 'humanize' yeast, the yNAA10 and yNAA15 loci were replaced by the corresponding human cDNA. First, hNaa10 WT or S37P was amplified from pBEVY-U-hNAA15-hNAA10  with primers adding homology to the yNAA10 locus (Table S3 in the Supporting Information: ON41/ON42: 5′-GGG AAA CCT AAA TAC ATA CGA TCA AGC TCC AAA ATA AAA CTT CGT CAA CCA TGA ACA TCC GCA ATG CGA GGC C and 5′-TGT GAA GAA GCC TGG ATG AAA ATA TAC TAC GTT TAT ATA GGT TGA TTT AAT CAG GAG GCT GAG TCG GAG GCC). Similarly to above, the PCR product was transformed into strain YG10 (yNaa10Δ::URA3-15) and transformants were selected using 5-FOA. To replace the yNAA15, hNaa15 cDNA was amplified from pBEVY-U-hNAA15-hNAA10 using ON54/ON55 (5′-CGT ACG ATA TCA TGC CGG CCG TGA GCC TCC CGC CCA A and 5′-GGC TAG ATA TCT CAA ATT TCA TTG GCC AGT TCT TCA GC) and cloned into pUG6-tTA (Zilio et al., 2012) to generate an hNaa15-KanMX cassette. The hNaa15-KanMX cassette was then amplified using ON59/ON59 (5′-GGA AGG CGA TTG ACC CTA ACG AAG TAT GCC GGC CGT GAG CCT C and 5′-AAT TGA CAC ATT GAG GAG TTG CAG GAA GCT CCT CGA GCG TCG ACA) to add homology for the yNAA15 locus. Additional homology was added using ON60/ON61 (5′ CCT TGT TCA AGA CAA ATA CCA TTG AGG AAG GCG ATT GAC CCT AAC and 5′ TAT ATA CAT AAA TTA AGT AAG AGT TAA TTG ACA CAT TGA GGA GTT). The PCR product was transformed into the respective yNaa10: hNaa10 strain and transformants were identified by selection on kanamycin plates. All strains generated in this study were validated by PCR and/or Sanger sequencing. For PCR verification, primers were designed to bind outside the intended sites of recombination such that a slower migrating product would be observed in the case of a correctly targeted transformation ( Table S3 in the Supporting Information: ON87/ON48). For Sanger sequencing verification, we confirmed that the genomic region immediately adjacent to the inserted transformation product corresponded to the SGD W303 reference sequence of the intended insert site. Further, we observed the absence of yArd1 and yNat1 transcripts in the humanized strains (yNatA::hNatA WT and yNatA::hNatA S37P) by RT-PCR (Fig. 1B). Briefly, the indicated S. cerevisiae strains were harvested during midexponential phase growth in YPD, and total RNA was extracted using the RNeasy Midi Kit (Qiagen). Following RNA extraction, mRNA was reverse transcribed to cDNA using Superscript III and a poly(T) primer as per the manufacturer's protocol. The resulting cDNA was amplified using PCR with primers designed to hybridize to regions internal to the indicated transcripts (Table S3:  . To control for the presence of genomic DNA, PCR reactions were run in parallel with an aliquot of the reverse transcription reaction that was run in the absence of Superscript (Fig. 1B, right side). In addition, expression of human and yeast Naa10 and Naa15 in the respective strains was verified by western blot using antibodies specific to human and yeast Naa10 and Naa15 (see Fig. 1C-E). A list of all strains used is shown in Table 1.

Growth assay in liquid culture
A single colony was picked and grown in 5 mL YPDA media (Clontech; endogenous strains) or SC -URA (Clontech, overexpressing strains) at 30°C overnight. The culture was diluted, grown for two further doublings, diluted to an OD 600 of 0.1 and incubated at 30 or 39°C for 24 h. The OD 600 was 22 M. J. Dörfel et al.

Stress test on solid media
Similar to the stress tests in liquid culture, a single colony was grown in 5 mL of the corresponding media overnight, then diluted to an OD 600 of 0.1. Subsequently, serial 1:5 dilutions were prepared and 10 μL of these spotted manually with a pipettor on YPDA or SC -URA plates. All plates were incubated for 24-48 h at 30 or 39°C to induce heat shock. For the chemical treatments, cells were spotted on plates containing 0-0.8 μg/mL cycloheximide or 0-400 μg/mL doxorubicin. For the UV treatment, plates were irradiated with 0, 15 and >20 mJ in a UV Stratalinker 2400 (Stratagene) after the cells were spotted. All experiments were performed at least three times.

Mass spectrometry
For protein expression analyses yeast were harvested during the exponential phase of growth at an OD 600 of 0.6, separated by filtration (see below) and cryogenically grinded in a SPEX Freezer Mill. The samples were resuspended in Y-PER yeast protein extraction reagent (Thermo Fisher) and cellular debris removed by centrifugation. Each sample was chemically labeled separately with one of up to eight distinct isobaric iTRAQ reagents (Ross et al., 2004). Peak lists were extracted using the MASCOT Distiller software (MatrixScience). False discovery rates (FDR) were estimated by searching equivalent reversed or randomized sequence databases, using a cutoff value of 1% FDR. Protein-level expression ratios were calculated using intensity-weighted averages from matching peptides after outlier removal. This experiment was performed in duplicate. Comparing the two sample replicates, we fitted a Cauchy distribution using the EasyFit statistical package (MathWave, Inc.). Peptide p-values were calculated using a two-tailed t-test.

Microarray
For gene expression analysis, RNA was extracted from a 10 mL yeast culture growing at exponential phase (OD 600 of 0.6) at 30°C using phenolchloroform extraction method. Briefly, cells were harvested by filtration through a 25 mm Nylon Hydrophilic Membrane Filter (Whatman), resuspended in 750 μL lysis buffer (10 mM EDTA; 0.5% SDS; 10 mM Tris, pH 7.5), mixed with an equal volume of saturated phenol, pH 4.3 (Fisher Scientific) and incubated at 65°C for 60 min with vortexing every 20 min. The samples were incubated on ice for 10 min, centrifugated for 5 min at 5000 g and the supernatant transferred to 2 mL phase lock gel tubes heavy and mixed with 750 μL chloroform. The aqueous phase was collected according to the manufacturer's instructions. The RNA was precipitated with 75 μL 3 M sodium acetate and 1.5 mL ethanol for 60 min at À20°C, pelleted at 3000 rpm for 10 min, washed twice with 70% ethanol and air dried at room temperature for 30 min. The RNA was dissolved in 25 μL H2O at room temperature, further purified using a RNeasy Mini Kit (Qiagen) and labelled with a Quick Amp Labeling Kit (Agilent Technologies) using cyanine-3-CTP/cyanine-5-CTP (Fisher Scientific). The probes were hybridized to 8 × 15 k yeast microarrays and scanned using a 2 μm Agilent microarray scanner. The images were analysed using the feature extraction software and normalized by global lowess using Goulphar. For each gene, the Cy5/Cy3 ratios corresponding to the duplicated probes were averaged. The averaged log2 ratios and the standard deviation between the two replicates were calculated for each gene.

Mating assay
The mating efficiency was assessed in a quantitative mating assay. Mating tester cells and cells of interest were grown in log phase to an OD 600 of 0.25. To generate tester plates, 0.1 mL of the tester strain was seeded on SC -Lys, -Ura selective plates and dried for 20 min. For each strain of interest, four 1:10 dilutions were prepared and 0.1 mL seeded on top of the tester plates. To determine the initial seeding number, 70 μL were seeded on SC -Ura selective plates. Cell colonies were counted after 2-3 days incubation at 30°C and the mating efficiency determined as the ratio between the numbers of colonies grown on the tester plates and initial number of seeded cells. The experiment was performed nine times.

Results
We previously identified a disease, Ogden syndrome, associated with a S37P variant in the human N α -terminal acetyltransferase, Naa10 (Rope et al., 2011). To study the impact of this mutation in vivo, we have studied different S. cerevisiae Naa10 variant stains and characterized the consequences induced by this mutation under various stress conditions.

The S37P mutation renders Naa10 functionally impaired
It was previously shown that disruption of Naa10 or Naa15 increases the sensitivity of yeast cells towards heat shock (Whiteway & Szostak, 1985;Mullen et al., 1989;. To test whether the Ogden mutation functionally impairs Naa10 in yeast, we performed growth experiments at an elevated temperature. In line with previous results, all tested knockout strains showed reduced growth at 39°C (see Fig. 2). The replacement of the yeast NatA complex with the human genes on the endogenous locus did not rescue the observed effects seen in the knockout (p = 0.195 for yNatA::hNatA WT and p = 0.595 for yNatA:: hNatA S37P) and already exhibited a slight but statistically significant (p < 0.01) growth defect at 30°C (Fig. 2A). Overexpression of the human NatA complex on the other hand complemented the strong defect to WT levels (p = 0.542 for yNatA WT and hNatA WT; Fig. 2B). The growth of the Ogden strain was significantly different to both the WT and the knockout (p = 0.005 and p = 0.001, respectively) indicating that overexpressing the mutated hNaa10 S37P/hNaa15 only partially restored the growth defect (Fig. 2B). To test whether the Ogden mutation has a similar effect on the function of the yeast Naa10, we first identified the homologous position using sequence alignment (Fig. 1A) of human and S. cerevisiae Naa10. Mutation of this position (yS39P) did not result in a heat shock phenotype (p = 0.184; Fig. 2A).
To confirm these results, stress tests on plates were performed. Similarly to the experiments in liquid culture, the knockout of the yeast NatA complex strongly reduced the ability of the cells to grow at elevated temperatures and this effect could not be rescued by replacing the yeast loci with the human NatA genes (Fig. 3A). Overexpressing the human NatA proteins on the other hand partially rescued the heat shock, with the mutant S37P being less efficient (Fig. 4A).
In addition to the heat stress, cycloheximide, doxorubicin and UV irradiation were used as stress factors. Cycloheximide blocks the translocation step in elongation and was used to challenge protein biosynthesis. Doxorubicin is an agent that intercalates into DNA, thereby trapping topoisomerase II covalently bound to DNA, which subsequently induces double-strand breaks (Patel et al., 1997). Similarly, elevated doses of UV radiation induce DNA-damage response pathways. Both were used since Naa10 has been previously connected to the regulation of cell survival in response to DNA-damage (for a recent review see (Dörfel & Lyon, 2015).
In analogy to the heat stress experiments, knockout of yNatA strongly affected growth on cycloheximide plates (Figs. 3B and 4A). This growth defect could partially be rescued by overexpressing the human hNat15/hNaa10 complex from plasmids andin contrast to the heat shockeven by expressing the human proteins from the endogenous yeast locus (Fig. 3B). The homologue yS39P mutation of the yeast Naa10 showed a similar growth pattern to the WT strain.
In the challenge with doxorubicin, knockout of yNatA clearly increased survival towards doxorubicin stress whereas overexpression of the human NatA complex or the human S37P mutant complex seemed to decrease growth compared with WT (Fig. 4B). No obvious differences in the phenotype could be observed between strains after UV- Deletion of NatA is associated with a dysregulation of heat shock proteins As discussed above, NTA regulates specific protein function, stability and expression. Therefore, we asked whether the phenotypic effects observed might be related to changes in the abundance of one or several proteins. To analyse the expression changes induced by the knockout or overexpression of NatA, we performed quantitative mass spectrometric analyses. In total, 3191 proteins were identified from which 491 were significantly dysregulated (fold changes >1.37 or <0.65 significant to >2 SD; p = 0.05) (Table S1 in the Supporting Information). A GO-term enrichment   (Wang et al., 2013;Zhang et al., 2005) of proteins with significantly elevated protein levels in the knockout strain (271 proteins) did not give any obviously interpretable results, except biosynthetic and metabolic processes (not shown). Proteins that were significantly less abundant in the knockout (220 proteins) were enriched for energy metabolism processes such as 'ATP synthesis coupled electron transport' (GO:0042773, adjp = 0.0019) or 'cellular respiration' (GO:0045333, adjp = 0.0033) and matingrelated processes ('adaptation of signaling pathway by response to pheromone involved in conjugation with cellular fusion', GO: 0000754, adjp = 0.0180). Furthermore, proteins that were dysregulated in the knockout by at least 25% but whose expression could be rescued by expression of the human NatA to within 10% of the WT level (172 proteins) were enriched for protein folding (GO: 0006457, adjp = 0.0191) and protein refolding (GO: 0042026, p = 0.0705). Protein folding and re-folding are accomplished by molecular chaperones, and from studies with Sup35/[PSI + ] prion strains, it is known that loss of NatA increases heat-shock response and increases Hsp104, Ssa1/2, Ssb1/2 and Sis1 protein levels (Holmes et al., 2014). In the analysis performed here, knockout of NatA was associated with a strong increase in the Hsp70 family proteins Ssa1-4 as well as Hsp104 and Sis1, and their expression was restored upon overexpression of hNatA WT (Fig. 5).
The Hsp70 proteins Ssb1 and Ssb2 were not found to be changed. Overexpression of the hNatA S37P mutant failed to restore the expression of the Ssa1-4, Hsp104 and Sis1 proteins. Additionally, Ssb1 and 2 were slightly less abundant in this overexpressing strain, as compared with the WT strain.
The knockout of Naa10/NatA induces a pseudo-diploid gene expression We next questioned whether the Naa10 knockout and the Ogden mutation might have an effect on gene expression. To address this, we performed gene expression analyses using microarray and RNA-Seq. Three different sets of yeast strains were used in the microarray study that were analysed jointly to capture biological variability: Set 1 included a W303 WT strain, a yNaa10 knockout strain and a strain where the endogenous locus of the yeast genes were replaced with human NAA15 and NAA10 WT or NAA10 S37P, respectively (see Table 1: strains YG2, YG1, YG42 and YG43). Set 2 was identical to Set 1 except that, instead of the single yNaa10 knockout, a double yNaa10/yNaa15 knockout was used (strains YG2, YG67, YG42 and YG43). Set 3 consisted of the overexpressing strains (strains YG33, YG35, YG32 and YG34). For RNA-seq, two biological replicates of the overexpressing strains were used (YG33, YG35, YG32 and YG34).
Figure 6(A) shows the gene expression levels for the analysed NATs from the microarray study, Figure 5. Mass spectrometric analyses of the overexpressing strains. Cells growing at exponential phase were harvested and lysed and the clarified lysates were subjected to quantitative mass spectrometry. The geometric mean of the measured abundances (relative expression) and the geometric standard deviation for each protein was calculated from two independent experiments and normalized. Shown is the difference of the expression compared with the WT strain for the HSP70 chaperones Ssa1-4, Ssb1-2 as well as HSP104 and Sis1

28
M. J. Dörfel et al. confirming the knockout of Naa10 and/or Naa15 in the corresponding strains. Also, the data shows that the deletions did not affect the expression levels of the other NATs. A table of the differentially expressed genes from the RNA-seq experiment is shown in Table S2 in the Supporting Information. In Fig. 7A (A) a top 10 list is shown of the genes that are down-or upregulated in all knockout strains. This also revealed that only the overexpression of the human proteins from plasmids could partially rescue the effects seen in the knockout whereas the expression from the endogenous locus did not. To quantify this effect, we looked at the genes there were consistently deregulated in the knockout of all three sets [log2(FC) < À1 or log2(FC) > 1, 39 genes]. Then we calculated the average fold expression change for the up-and downregulated genes in each condition for the endogenous sets and the overexpression set. These genes were less deregulated upon overexpression of the human NatA complex, illustrated by a reduction of the average deviation from the WT expression levels (Fig. 6B). In contrast to this, when the human complex was expressed from the endogenous locus, the average expression change compared with WT was similar to that seen upon knockout of yNatA, indicating no rescue effect. Among the genes that are downregulated in the knockout strains of all sets by log2 FC ≤ À0.5 (117 genes) a GO term analysis using WebGestalt revealed enrichments for mating and fusion. The top hits were 'cell surface receptor signaling pathway' (GO: 0007166, adjp = 7.68 × 10 À7 ), 'pheromone-dependent signal transduction involved in conjugation with cellular fusion' (GO:0000750, adjp = 7.68 × 10 À7 ) and 'response to pheromone' (GO:0019236, adjp = 7.68 × 10 À7 ). Similar results were obtained when using the RNA-seq data (not shown). Early studies showed that depletion of Naa10 results in the derepression of the normally silent α-information at the HML locus (Whiteway & Szostak, 1985;Whiteway et al., 1987). Therefore, we speculated that the induced expression of the alpha1/2 proteins would radically change the expression profile in ΔNatA MATa cells. Indeed, a detailed analysis of mating-type specific genes showed a downregulation of a-specific genes such as the receptor for alpha pheromone, Ste2, and haploid-specific genes (e.g. HO) in the yNaa10/NatA knockout cells (Fig. 7B). and downregulated (red) genes for the endogenous rescue strains (left) and the overexpression strains (right). The darker shades show the results for when all genes were plotted, whereas the lighter shades show only those genes whose expression was changed by at least AE1 in all knockout strains. Wilcoxon rank sum test was used to analyse whether the expression change was significantly reduced in the rescue strain when compared with the corresponding knockout. ** p < 0.001; * p < 0.05 A downregulation of α-specific genes in the MATa cells was not observed. A similar silencing mechanism has been described for the telomeric regions (for a review see Gartenberg 2000). Since we expected that a disruption of telomeric silencing would increase the expression of genes specifically in that region, we plotted the relative position of the most upregulated genes (see also Fig. S1 in the Supporting Information) in the knockout according to their relative position on the chromosomes. As shown in Fig. 7C for the overexpressing strains, these genes cluster towards the end of the chromosomes, indeed indicating a de-repression of subtelomeric regions. This effect could also be observed in the endogenous set, albeit to a lesser extent (not shown). Overexpression of the human NatA complex did not restore the repression of these genes (Fig. S1 in the Supporting Information). A similar analysis of telomeric region expression was also performed using RNAseq data. By considering log2 fold change > 1 and Benjamini-Hochberg adjusted p-values <0.05, 69 genes were found to be upregulated in the knockout cells, relative to the wild type (Fig. 7D). Consistent with microarray data, we found that upregulated genes are significantly enriched in the telomeric regions (p-value = 0.04, Chi-square test). Together, these results suggest that the disruption of telomeric silencing owing to knock out of the yeast NatA complex increases the gene expression levels in those regions.

Knockout of Naa10/NatA affects mating
The pseudo-diploid gene expression pattern in MATa Naa10/NatA knockout yeast should interfere with the ability of the cells to mate. To test this hypothesis, we performed quantitative mating experiments. Indeed, the knockout of the NatA complex completely abolished mating in yeast (Fig. 8). The overexpression of the human NatA complex (hNatA) rescued the mating ability of the cells to WT levels; however, overexpression of the mutated Naa10 S37P/Naa15 complex showed a statistically significant diminished rescue of mating when compared with the WT strain.

Discussion
Here we show that knockout of the yeast NatA complex induced a strong growth defect at elevated temperatures in liquid culture and on plates. This is in line with previous findings that knockout of Naa10 or Naa15 induces heat sensitivity in yeast. Whiteway & Szostak, 1985;Mullen et al., 1989;. Overexpression of the human NatA complex rescued this growth defect at 39°C whereas the mutated hNaa10 S37P/hNaa15 only partially restored the defect, indicating a functional impairment of the Ogden mutation in vivo. The reason for the heat sensitivity is not known to date; however, GO term enrichment analyses from our proteomic characterization suggest that disruption of NatA might perturb protein folding and re-folding, which could contribute to this phenotype. Furthermore, a recent study showed that disruption of NatA in a [PSI + ] prion strain led to the accumulation of protein aggregates and the activation of heat-shock response proteins of the Hsp104, Hsp70 and Hsp40 family (Holmes et al., 2014). Similarly, in our mass spectrometric analyses, Hsp104, Ssa1-4 (Hsp70) and Sis1 (Hsp40) were increased in the knockout under normal growth conditions, whereas they were back to control levels upon overexpression of hNatA.
Overexpressing of the mutated hNatA S37P complex only partially compensated for this effect, which is in line with the partial rescue observed in the growth tests. This finding indicates that the observed effects may be caused by an impaired functionality of these chaperones, possibly owing to a reduced acetylation of their N-termini. The upregulation of these proteins could be a naturally occurring mechanism of the cells to compensate for their diminished activity. Indeed, Ssa1-4 and Ssb1/2 proteins have been shown to be NatA substrates  and loss of NTA of Ssa1/2 and Ssb1/2 has been shown to disturb their functionality in the propagation of the prion phenotype (Holmes et al., 2014). Additionally, overexpression of the heat-inducible Hsp70 proteins Ssb1/2 could partially suppress the temperature sensitivity in Figure 8. Quantitative mating experiments. To quantify mating efficiencies, cells were seeded on SC -Lys,-Ura selective plates containing a lawn of mating tester cells and in parallel on SC -Ura plates to determine the seeding cell number. Cell colonies were counted after 2-3 days and the mating efficiency was determined by calculating the ratio between the mated and the seeded cells. Shown is a box plot representation for the results from nine independent experiments. Statistical analyses were performed using Student's t-test ΔyNaa15 or ΔyNaa10 cells (Gautschi et al., 2003). This suggests that loss of NTA of the Hsp70 chaperones Ssa1-4 and Ssb1/2 in NatA cells reduces the functionality of these proteins, thereby affecting protein biogenesis and reducing cellular fitness under normal conditions as seen in the liquid growth experiments. Similarly, primary fibroblasts from the Ogden patient exhibit a decreased growth rate and proliferation when compared with control cells (Myklebust et al., 2015). Under stress conditions, when there is a heavy load on protein folding, this effect could be further aggravated, as seen in the heat shock experiments presented here. It would be interesting to see if an elimination of NTA from Ssa1/4 and Ssb1/2 is sufficient to reproduce the heat sensitivity or if the defect of ΔNatA cells could be restored to WT levels by artificially acetylating these proteins. The latter could be done in future experiments by mutating the second (first after iMet excision) amino acid of these proteins to change their NAT specificity and/or to prevent their NTA entirely, similarly to the experiments performed for the propagation of the prion phenotype (Holmes et al., 2014). In analogy to the temperature challenge, similar results were obtained when cycloheximide was used as chemical stressor. Knockout of yNatA strongly decreased the resistance of yeast towards cycloheximide and this defect could be partially rescued by overexpression of the human hNat15/hNaa10 complex from plasmids. The hNaa10 S37P mutant showed a similar or slightly decreased rescue effect, when compared with hNaa10 WT. This is in line with previous experiments where the sensitivity of ΔyNatA yeast towards cycloheximide and caffeine was nearly completely repressed by the expression of hNatA, whereas expression of hNatA S37P resulted in a reduced complementation .
A similar trend could be observed when we used doxorubicin as a stressor; however, in this case Naa10 negatively regulates cell survival. In fact, knockout of yNatA improved survival of the cells, whereas overexpression of the hNatA reduced cell growth on doxorubicin-containing plates. From studies in Drosophila melanogaster and HeLa, Ht1080 and U2OS cells, it is known that siRNA-mediated knockdown of Naa10 protects the cells from doxorubicin-induced cell death (Yi et al., 2007;Yi et al., 2011), which indicates that Naa10 has a regulatory function in DNA damage control. Indeed, doxorubicin treatment induced RIP1/PIDD/NEMO complex formation, NEMO ubiquitination and NF-κB activation in HEK293 (Park et al., 2012), providing a possible mechanism for Naa10 function in DNA damage. However, the NF-κB pathway is absent in yeast (Srinivasan et al., 2010), so the mechanism in yeast needs further study. Similarly to the effects observed for doxorubicin, knockdown of Naa10 also increased resistance of HeLa cells towards cisplatin and ultraviolet treatment (Yi et al., 2011). In contrast to this, we did not observe any growth or survival advantage of a NatA knockout in yeast under conditions of UV irradiation.
The replacement of the yeast NatA complex with the human genes on the endogenous locus did not reproducibly rescue the heat shock defects seen in the knockout, along with more variability noted among the different biological replicates (Fig. 2). One explanation for this finding is that the human proteins might not be expressed as efficiently as the corresponding yeast proteins in S. cerevisiae (also see Fig. 1). Furthermore, despite the NatA complex being evolutionarily conserved from yeast to vertebrates (Sugiura et al., 2003;Arnesen et al., 2005), with all tested variants having a similar substrate specificity, slight structural differences could result in the human proteins being less functional in yeast. A comparative N-terminal acetylome analysis using COFRADIC revealed that, although human and yeast NatA acetylate the same set of proteins, hNatA displays a preference towards Ala-N-termini whereas yNatA seems to be more efficient in acetylating Ser-starting N-termini . The latter study showed that yNatA seems to be more efficient than the human complex in yeast . Hence, the expression of the human proteins at the endogenous locus might not be sufficient to complement yeast NatA, whereas overexpression from a plasmid could make up for a diminished acetylation activity of the human proteins towards yeast substrates. Similarly, overexpression of the human proteins rather than endogenous expression was necessary to revert the expression changes induced by disruption of yNatA. Endogenous expression of the human proteins partially rescued a growth defect towards cycloheximide treatment, 32 M. J. Dörfel et al. which could indicate that NatA substrates involved in the cycloheximide response are less susceptible to a reduction in NTA. In contrast to the effects observed with the hNatA S37P mutation, the homologous mutation in yeast Naa10 showed no observable effects. This suggests that there are significant structural differences between the species. In this regard it has been shown that heterologous combinations of human and yeast Naa10/Naa15 are not functional in yeast . Also, the sequence alignment performed in this study showed an additional putative α-helix C-terminally adjacent to the S39 in S. cerevisiae, which is missing in the human Naa10. Furthermore, the Ogden mutation is not completely deleterious in human Naa10 since residual catalytic activity of Naa10 S37P could be found in vitro (Rope et al., 2011), in a yeast model  and in primary cells from the patients (Myklebust et al., 2015). Therefore, structural difference and the relatively mild nature of the mutation itself could explain the absence of an overt phenotype in yeast.
Taken together, the data presented here indicates that Naa10 positively regulates cell survival in response to cycloheximide and heat stress but promotes cell death in response to doxorubicin. The human NatA proteins could rescue the yeast knockout, at least when overexpressed. The Naa10 S37P mutation was found to be less efficient than wild-type human Naa10 for the phenotypes involving heat shock, chaperone protein accumulation and mating, indicating that the Ogden mutation is affecting Naa10 function in vivo for at least a few phenotypes, supporting the disruptive nature of this mutation.
To analyse the effects of NatA disruption at the transcriptional level, we performed microarray studies on 3 different sets, each containing a WT strain, a Naa10/NatA knockout and 2 rescue strains expressing the human NatA WT or S37P proteins, respectively. A GO-term enrichment analysis of the downregulated genes in the knockout revealed that these genes were involved in mating, cellular fusion and response to pheromone. We also found a similar enrichment in the mass spectrometric analyses, highlighting that the downregulation of genes involved in mating/fusion can also be detected at the protein level.
In S. cerevisiae, mating is regulated by a-and αspecific transcription factors that are encoded by the nonhomologous alleles MATa and MATα (Haber, 2012). The MAT locus is encompassed by two silenced alleles, HML and HMR, that contain a cryptic copy of the mating type sequences MATa and MATα, respectively, that serve as donors for the MAT locus to confer the mating type of a haploid cell (Haber, 2012). Silencing of the HML and HMR is accomplished by trans-acting factors that bind to specific silencer sequences surrounding these loci. First, the origin recognition complex ORC binds to the silencer and recruits silent information regulator, Sir1 (Fox et al., 1997). This recruits the Sir2/Sir3/Sir4 complex which deacetylates the tails of histone H3 and H4, thereby creating additional binding sites for Sir3/Sir4, resulting in a sequential spreading of these proteins and transcriptional repression in this region (Rusche et al., 2003). Sir3 and Orc1, the large subunit of the origin recognition complex, are both substrates of NatA, and NTA of these proteins stabilizes their binding to nucleosomes, thereby facilitating silencing of HML/HMR (Wang et al., 2004;Arnaudo et al., 2013;Geissenhöner et al., 2004;van Welsem et al., 2008). Disruption of NatA should therefore induce the expression of a-and α-genes from the de-repressed HML/HMR loci, leading to a pseudo-diploid gene expression profile. Our microarray and RNA-seq data supports this hypothesis: a-and haploid-specific genes were downregulated in ΔyNaa10/yNatA knockout cells, when compared with the WT. In haploid a-cells, as used in this study, alpha-specific genes are constitutively repressed. Diploid cells contain both MATa and MATα, which results in the formation of the a/alpha2 repressor that keeps alpha-specific genes repressed (Goutte & Johnson, 1988). In analogy to this, the artificial expression of HML/HMR genes as a result of the yNaa10/yNatA knockout did not affect the gene expression of alpha-specific genes in our analysis, instead they remained repressed. This pseudodiploid expression profile (e.g. downregulation of the receptor for alpha-factor, Ste2 or repression of Ste4, a G protein involved in the mating signaling pathway) was accompanied by a dramatically reduced mating efficiency as shown in quantitative mating experiments. This is in line with previous findings that disruption of NatA interferes with mating (Whiteway & Szostak, 1985;Whiteway et al., 1987;Aparicio et al., 1991). This mating effect is seen only in MATa cells, 33 A yeast model for Ogden syndrome as used in this study, since silencing at HML is less stable than that at HMR, especially when Sir3 becomes limiting (Geissenhöner et al., 2004;Motwani et al., 2012). In agreement with the stress tests, the pseudo-diploid gene expression pattern as well as the mating efficiency could be partially reversed by overexpression of the human NatA complex. Furthermore, overexpression of the Ogden mutant was less efficient in rescuing the mating defect.
Sir3 is also involved in silencing at telomeres (Geissenhöner et al., 2004;Ruault et al., 2011). An impaired functionality of this protein as a result of their reduced NTA should therefore lead to derepression of genes in that region. Consistent with this, we could detect an upregulation of genes in sub-telomeric regions in the NatA knockout strains by RNA-seq and microarray. Also, in contrast to the pseudo-diploid expression profile, the derepression of genes at telomeres could not be complemented by overexpression of the human NatA proteins. It is known that telomeric silencing is less robust than HML/HMR silencing (Haber, 2012) and might therefore be more susceptible to small perturbations. Since the human NatA proteins seemed to be less functional towards yeast proteins (see above), it could be possible that the overexpression of hNatA only partially restored NTA of Sir3 and Orc1, which in turn might be sufficient to restore HML/HMR but not telomeric silencing.
In conclusion, we have shown here that disruption of NatA leads to growth defects and increased sensitivity towards stresses such as heat or cycloheximide in S. cerevisiae. These effects are possibly due to an impaired functionality of molecular chaperones. Additionally, ΔNatA cells are characterized by a pseudo-diploid gene expression pattern and failure to mate, probably owing to reduced NTA of the NatA substrates Orc1 and Sir3. These effects could be restored by expression of the human NatA complex to different degrees, depending on the system used (endogenous expression vs. overexpression) and other cellular processes. A mutated form of hNaa10, harbouring the variant implicated in Ogden syndrome, displayed a similar or diminished complementation when compared with WT hNaa10. This strongly supports the disruptive nature of the Ogden/S37P mutation in Naa10, consistent with prior results (Myklebust et al., 2015;.

Data deposition
The microarray and RNAseq data was deposited in GEO under accession number GSE86482. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (Vizcaino et al., 2016) partner repository with the dataset identifier PXD004923.

Supporting information
Additional supporting information may be found in the online version of this article at the publisher's web site: Figure S1. Heat map representation of the microarray resultsfor the genes that were found to be upregulated by at least 3-fold in the yNata knockout (overexpressing strains only). Table S1  Table S2  Table S3